Our Mission

Building the Autopilot
for commercial HVAC

Noviterm AI is a DeepTech startup bridging the gap between legacy building infrastructure and advanced artificial intelligence — using Physics-Informed Neural ODEs and Model Predictive Control to deliver measurable energy savings with zero operational risk.

20–40%

energy cost reduction per building

~45 t

CO₂ avoided per building / year

Day 1

no historical BMS data required

12–18

months typical payback (ROI)

What we believe

Decarbonize legacy commercial real estate through
applied mathematics — not guesswork.

Our mission is simple: deliver measurable energy savings in every commercial building we touch — using physics, not black-box statistics. We cross-validate every Digital Twin against EnergyPlus and OpenModelica before deployment, so the savings you’re promised are the savings you get.

Physics first

Neural ODE models thermodynamic equations, not statistical patterns. R² = 0.991 validated accuracy.

Privacy by design

All AI runs on-premises. Your building data never leaves the facility — EU data sovereignty compliant.

Verified savings

Every kWh measured against IPMVP Option A/B baseline. Automated ESG / CSRD / ISO 50001 reports.

Leadership

The people behind
the physics

Anton Kostiuk

Anton Kostiuk

Founder · CEO & CTO

15+ years in HVAC system design and B2B implementation. 300+ commercial HVAC & automation projects. Creator of the Neural ODE + MPC core algorithm.

LinkedIn →

Oleksii Kuzmenko

Oleksii Kuzmenko

Frontend & Web Engineer

Atomic design systems · Component Journey Mapping · UI Kit architecture · Grafana dashboards. Turns complex energy data into clear, actionable interfaces.

LinkedIn →

Dmytro Babenko

Dmytro Babenko

HVAC Domain Lead

Thermal and Automation engineer · 20+ years in thermotechnics, HVAC automation & system design · 250+ commercial projects across Ukraine and Europe.

Backed & supported by

Google for Startups, Greencubator, Ukrainian Startup Fund, UNIDO

Company Details

Our vision

Every building running on AI autopilot by 2035

75% of EU commercial buildings are energy inefficient. We’re building the software layer that makes them smart — without replacing any hardware, without months of preparation.